Boosting in Supervised Machine Learning by Mohammad Mehdi Mojarradi ’23, Statistics Colloquium

Boosting in Supervised Machine Learning by Mohammad Mehdi Mojarradi ’23, Statistics Colloquium, Wednesday, April 12, 1:10 – 1:50 pm, North Science Building 114, Wachenheim.

Abstract:  Statistics is a field famed for perplexing results backed by mysterious mathematics (see: central limit theorem, stein’s paradox).  The method of boosting contributes to this reputation.

Used to solve classification problems in supervised machine learning, boosting works by sequentially applying a classification algorithm to reweighted versions of the training data, and then taking a weighted average of the sequence of classifiers thus produced.  This simple strategy can dramatically improve performance.  The mathematical justification for why it works, however, is not so simple, and explaining it will be the subject of my talk.  The discussion will follow the important work of Friedman, Hastie, and Tibshirani (2000) in “Additive Logistic Regression: A Statistical View of Boosting.”